Describir: Parallel processing area extraction and data transfer number reduction for automatic GPU offloading of IoT applications